Psychiatric Disorders in America: the Epidemiologic Catchment Area Study

Psychiatric Disorders in America: the Epidemiologic Catchment Area Study

Lee N. Robins, Darrell N. Regier

New York, NY: The Free Press, 449 pp., 1991

The President’s Commission on Mental Health was charged to examine U.S. needs for mental health services and the need for new knowledge about disorders. Freedman (1978) noted the Commission’s concern with the public need for a sound knowledge base for planning services and use of human resources. He also stated that “the Commissioners perceived the vulnerability of the mentally ill to special neglect” as well as “becoming vehicles for pet social slogans.” The Commission required knowledge of the range and magnitude of serious psychological problems in the population in order to estimate the burden of illness, the potential for prevention, and the identification of high risk groups for the purpose of allocating resources for research, services and personnel training.

As part of the Commission’s work a book was published (Dohrenwend et al 1980) summarizing the state of knowledge up to that time. This served to highlight a number of problems in the existing estimates. There was a wide variation in the estimates of current prevalence. The diagnostic and classification systems in use lacked consistency, reliability and validity. Case ascertainment methods differed widely, as did case definitions. Sampling techniques also showed wide variation, though some of the studies since the 1950s used sophisticated sampling techniques, for example the Stirling County (Leighton et al 1963), Midtown Manhattan (Srole et al 1962), and New Haven (Weissman and Myers 1978a; Weissman and Myers 1978b; Weissman et al 1978; Weissman and Myers 1980) projects. Some studies concentrated on urban populations, some on rural and some on special situations such as Hutterite colonies (Eaton et al 1955), or Pacific Northwest Coastal Indians (Shore et al 1973).

To implement the requirements of the President’s Commission for epidemiological data, NIMH was faced with the task of producing an interview that incorporated the newer diagnostic concepts of DSM-III and was suitable for administration to large numbers of people at a reasonable cost. The Renard Diagnostic Interview (Helzer et al 1981), originally designed to operationalize the St. Louis’ criteria was chosen for redevelopment. This resulted in the Diagnostic Interview Schedule (DIS: Robins et al 1981a; 1981b) which could be administered by trained lay interviewers in about one hour, and yielded DSM-III diagnoses for major disorders which were generated by a computer program. Therefore the cost could be kept down by eliminating the use of trained professionals as interviewers, and the exercise of variable clinical judgment was eliminated by the use of computerized diagnoses.

Thus the study conceived in 1977 by NIMH with the impetus of the President’s Commission, and later known as the Epidemiologic Catchment Area Program (ECA), came into being. The results are reported in this book.

Five centers, New Haven, Baltimore, St. Louis, Durham and Los Angeles, became involved in the project, all with samples of over 3000 household resident subjects interviewed, for a total sample of about 20,000. At each site the samples were drawn from a population of over 250,000. Some cities had a considerable proportion who were black, and Los Angeles had a significant Hispanic population. Rural and urban residents were subsequently analyzed separately for many comparisons. The methods of obtaining random samples varied somewhat, including the use of property files, utility company listings and block sampling to select households. Within households respondents (age 18 and older) were selected using Kish grids to ensure a distribution similar to that of the population. Three sites developed methods for over-sampling the elderly. All sites developed methods for sampling those resident in institutions including, nursing homes and chronic hospitals, prisons and psychiatric hospitals. A total of 1379 response from institutional residents were obtained. The response rate from the community survey (household respondents) varied from 68% to 79%. All respondents were administered the DIS and a Health Services Questionnaire by trained lay interviewers. Remarkably, less than 1% of those who agreed to be interviewed failed to complete the interview. The DIS, though using DSM-III criteria, does not cover all the 122 adult diagnoses in DSM-III. It had to be confined to disorders for which criteria were clear and which could be determined from interview alone. Also, to keep the interview to a tolerable length, only those disorders deemed most important were included. Those reported and included in the book were: affective disorders (major depression, dysthymia, bipolar, atypical bipolar), schizophrenia and schizophreniform disorders, a variety of alcohol and substance use disorders, anxiety disorders (obsessive compulsive, agoraphobia, social phobia, simple phobia, panic and generalized anxiety), somatization, antisocial personality disorder, and cognitive impairment (not a DSM-III diagnosis). Other disorders, such as post traumatic stress disorder and anorexia nervosa, are not reported in the book but are included in some of the other published papers.

The diagnostic programs used to analyze DIS data can be set to use all of the DSM-III criteria for a disorder (both inclusion and exclusion) or modified. For this book (and most of the published papers) the exclusion criteria of DSM-III have been ignored, where the exclusion would be based on a pre-existing or co-existing psychiatric disorder. DSM-III exclusion rules frequently lack specific details, are not always based on research evidence, do not give rules for deciding when one disorder is “due to” another and, if the rules were applied, would preclude examining the relationships between disorders. Thus when reading any paper using DIS interview data it is important to check what exclusion criteria are being used or whether they have been ignored. Regardless of which rules are used the diagnoses are categorical, that is present or absent.

Many subjects have symptoms (sometimes of more than one disorder) which may just fail to meet diagnostic criteria. Such subjects are not classed as “cases,” although they may have significant psychopathology. In community studies, there are many such borderline “cases.” Also, it is important to ensure that representative population prevalences be obtained rather than relying on hospital or outpatient data reports which may represent only a small proportion of cases possibly biased by self selection. Estimates of unmet needs cannot be obtained, if those with the unmet needs are not sampled.

Each of the book chapters detailing specific disorders offers comments and data on diagnosis and prevalence rates (often with age, sex, and race specific data). Some have details on income, social class, marital status, subtypes of disorder, employment, urban-rural differences, use of health services, age of onset, duration of disorder and comorbidity. Not all chapters are consistent, reflecting aspects that are of interest for particular disorders.

A review such as this cannot cover the wealth of detail presented in the book and only some major points can be mentioned. The President’s Commission estimated that 10 – 15% of the population had an active psychiatric disorder (one year prevalence). The EC A estimates are that 32% of the population have or have had a psychiatric disorder (lifetime prevalence), and that 20% had an active disorder (meaning that they had met criteria for a disorder at some time in the person’s life and had at least one symptom in the year prior to the interview). With the exception of cognitive impairment, between 52% and 100% of persons with a diagnosis had also met lifetime criteria for a least one other diagnosis. Even when co-occurrence within the last year only is considered, patterns are emerging which suggest that these findings are not an artifact of reporting style or other confounding factors.

Comorbidity is thus receiving increased attention. The concept of “one patient, one diagnosis” is no longer tenable. The basis for exclusion criteria must be questioned and dogma replaced by research. Studies of the effects of comorbidity on treatment and outcome are needed to provide guidelines for clinicians and more effective treatment for patients.

Psychiatric disorders typically begin at a young age. Of all those with a disorder, the median age at the first symptom was 16, and 90% have had their first symptom by age 38. Therefore, with the exception of cognitive impairment, those over age 40 add few later life onsets. In general, later age of onset was associated with higher rates of recovery. For those who had a recovery, the average duration from first to last symptom was 10.4 years.

It is well known that the incidence and prevalence of some disorders is quite different for males and females. Major depression is more common in women, but alcoholism and antisocial personality disorders are more common in men. Many earlier studies reported that more women than men had a psychiatric disorder over their lifetime. This may be merely a reflection of which disorders were included in the study. Since the EC A included antisocial personality and alcoholism it is not surprising that the lifetime prevalence for all disorders was higher in men (36%) than women (30%), but the one year prevalence (20%) was the same for both sexes.

A somewhat puzzling finding from the ECA is that the elderly showed low rates of both current and lifetime rates for psychiatric disorders. Although methodological issues have been raised to question whether this is an artifact, there are good arguments that it is a real finding. Explanations include the differentially high mortality of those with psychiatric disorders — thus those with a disorder do not survive to become elderly, and that there is an increasing rate of psychiatric disorders occurring in successive birth cohorts.

Rates of disorders varied by marital status, with those who had been separated or divorced having higher rates than the single, and those who had remained married, having the lowest rates. Those who were unemployed or had unskilled jobs had high prevalence rates.

Only 19% of those with a current disorder reported treatment within the last year, yet 4% of those who did not have a disorder covered by the DIS had been in treatment. This means that almost equal numbers of people with or without a diagnosis were being treated. Only 11% of those with antisocial personality and 15% with alcohol abuse had ever told a doctor about their problems, contrasted with 47% for schizophrenia, 61% for depression, and 73% for panic.

Those with multiple disorders were more likely to have been treated than those with a single disorder. Studies of treated cases show that depression and anxiety are the most commonly treated diagnoses, both being more common in women and more widely recognized as psychiatric disorders than substance use or antisocial personality. In contrast, the most common disorders from the community survey were (in descending order) phobia, alcohol abuse/dependence, generalized anxiety, major depressive episode and drug abuse/dependence. This holds whether lifetime or one year prevalence is being considered.

The ECA study has met its major objective, namely, to provide information about the number of people suffering from various mental disorders. It has also identified various underserved and high risk groups. The high proportion of people with disorders who go untreated has been measured. Many other items of importance have also emerged. The age at risk for the onset of disorders, relationships to sex, marital status, employment, education and social class have all been studied and new knowledge derived. The very small proportion of those with alcohol abuse/ dependence, antisocial personality and drug abuse/dependence who report symptoms to a physician, let alone get treatment, clearly shows defects in the delivery of service. It may be argued that treatment of these disorders leaves much to be desired, but the social consequences of these problems cannot be left forever unattended. Groups who are less likely to seek care such as married men are identified. Comorbidity is now recognized as an entity rather than an artifact of diagnostic ineptitude, and thus patients are likely to receive a comprehensive evaluation rather than just concentrating on presenting complaints. Some commonly held beliefs, eg., that women have higher rates of disorder than men, that vulnerability to disorder increases with age, and that anxiety and depression are the most common disorders, have been questioned or discarded. The low rates of disorders in the elderly, and the possibility of increasing rates in younger birth cohorts found in the ECA raise new questions.

Methodology has been developed which allows comparable studies to be conducted in other locations, and this is attested to by the many languages in which the DIS is now available. Efforts have already been made to examine similar data to compare rates and patterns for illnesses in different parts of the world. This era of internationally comparable studies has great potential.

At a different level, the tools and the benchmarks are now available to assess whether programs for prevention, treatment, outcome, and use of health services — including policy initiatives — really have the desired effects.

One of the most important results is that material is at hand for public education. This includes information on the frequency of disorders, symptoms and how to recognize them, the likelihood of someone seeking treatment, the duration of the illness, its age of onset, and the likelihood of recovery.

Changes in health services and their accessibility and availability are suggested when, for some treatable disorders, so few receive care. This book provides the most detailed and comprehensive information available on the range, distribution and frequency of psychiatric disorders. It has importance to clinical work, health promotion, prevention, organization of health and social services, and future research. The reader must remember, however, that the study relates to the U.S., and even in one country considerable intersite variation was found. Extrapolation of results to other settings must be done with caution. It is nevertheless one of the most important books to be published on psychiatric disorders. While it is replete with tables and detailed information, more detail may often be found in the other published papers and books from the ECA project.

Psychiatrists, residents, mental health workers and researchers will want access to this book, as will many members of the public, who now have sufficient information to ask the right questions.